Tesi etd-03182022-102821
Link copiato negli appunti
Tipo di tesi
Dottorato
Autore
GHERARDINI, MARTA
URN
etd-03182022-102821
Titolo
The myokinetic control interface for hand prostheses - Interface optimization for trans-radial amputations
and feasibility of application to TMR patients
Settore scientifico disciplinare
ING-IND/34
Corso di studi
Istituto di Biorobotica - BIOROBOTICS
Commissione
relatore Prof. CIPRIANI, CHRISTIAN
Parole chiave
- human-machine interface
- magnetic field.
- magnetic tracking
- myokinetic interface
- upper limb prosthetics
Data inizio appello
31/05/2022;
Disponibilità
parziale
Riassunto analitico
The quest for a human-machine-interface for the control of hand prostheses able to restore the functionality of the lost
limb in an intuitive and physiologically appropriate way has not been accomplished yet. In the last decade, much effort has
been dedicated to explore innovative control strategies, which mostly rely on the registration of the electrical signals generated by the muscles following nerve stimulation to decode the users’ intention of movement. In contrast to this, an innovative approach, dubbed the myokinetic control interface, aims at deriving the control signals by exploiting the muscle movements as control source, instead of their electrical activity. This can be achieved by implanting small permanent magnets inside the residual muscles of the stump, collecting the magnetic field they generate using external magnetic sensors, and
subsequently applying magnetic tracking strategies to monitor their displacement. Indeed, as the magnets would follow
the movement of the muscles they are implanted in, localizing them would allow to continuously determine the muscle
contraction state, and the latter could be translated into appropriate commands for an artificial hand.
The myokinetic control interface was first proposed in 2017 and, if successful, holds the promise to bring several advantages for people living with limb loss. Indeed, such interface could potentially provide a parallel control over multiple degrees of freedom of the prosthetic device in a way perceived as natural by the user and, since it is based on passive implants,
it would be intrinsically safe. In this perspective, the realization of a real-time, accurate, and reliable multi-magnet localiizer represents an imperative step for ensuring the success of the envisioned system. To this aim, the present dissertation is a collection of works aimed at unveiling the basic principles of the localization problem, providing concrete guidelines for the magnets implant, as well as defining an optimal system design for its practical implementation.
The myokinetic interface was originally conceived for addressing trans-radial amputations, because the extrinsic muscles
originally devoted to specific functions in the lost hand (e.g., the one responsible for flexing the fingers), and which are hosted inside the forearm, need to be targeted by the implant. This is because only by exploiting such native control sources we could realize an interface capable of restoring a natural control for the user. For this reason, after the search of general
guidelines in a simplified environment, the first part of the present dissertation is dedicated to the development and optimization of a multi-magnet localizer in a workspace resembling the human forearm, which is anatomically relevant for
the final application. By simulating different amputation levels, we determined the number of implantable magnets given the
stump length, as well as an optimal magnet placement inside the muscles that guarantees highly accurate localization outcomes. In addition, a sensor selection strategy aimed at reducing as much as possible the computational cost and power
consumption of the localization, while keeping the relevant information needed for localizing the magnets is presented.
On the other hand, the second part of this thesis tries to overcome the initial limit of application foreseen for the myokinetic approach, and to go beyond trans-radial amputations by addressing more proximal (above-the-elbow) ones, while
still preserving the expected naturalness of use of the proposed solution. This is made possible by the recent progress
in surgery, which has led to the so called targeted muscle reinnervation (TMR) technique: by redirecting the nerves that
served the lost limb to reinnervate surrogate muscles, such technique provides novel and physiologically appropriate control sources for artificial limbs. Thus, by merging the advances from surgery with those from technology, the viability and the
benefits of applying the myokinetic approach to proximal amputations in TMR patients were demonstrated. Specifically, experimental sessions carried out with two TMR patients proved the feasibility of applying the myokinetic interface, in a non
invasive way, to effectively decode different voluntary movements using a simple logistic regressor.
The outcomes of the present work not only are of interest for the development of a novel control interface for hand prostheses, but they also provide interesting insights for a wide range of bioengineering applications exploiting magnetic tracking.
limb in an intuitive and physiologically appropriate way has not been accomplished yet. In the last decade, much effort has
been dedicated to explore innovative control strategies, which mostly rely on the registration of the electrical signals generated by the muscles following nerve stimulation to decode the users’ intention of movement. In contrast to this, an innovative approach, dubbed the myokinetic control interface, aims at deriving the control signals by exploiting the muscle movements as control source, instead of their electrical activity. This can be achieved by implanting small permanent magnets inside the residual muscles of the stump, collecting the magnetic field they generate using external magnetic sensors, and
subsequently applying magnetic tracking strategies to monitor their displacement. Indeed, as the magnets would follow
the movement of the muscles they are implanted in, localizing them would allow to continuously determine the muscle
contraction state, and the latter could be translated into appropriate commands for an artificial hand.
The myokinetic control interface was first proposed in 2017 and, if successful, holds the promise to bring several advantages for people living with limb loss. Indeed, such interface could potentially provide a parallel control over multiple degrees of freedom of the prosthetic device in a way perceived as natural by the user and, since it is based on passive implants,
it would be intrinsically safe. In this perspective, the realization of a real-time, accurate, and reliable multi-magnet localiizer represents an imperative step for ensuring the success of the envisioned system. To this aim, the present dissertation is a collection of works aimed at unveiling the basic principles of the localization problem, providing concrete guidelines for the magnets implant, as well as defining an optimal system design for its practical implementation.
The myokinetic interface was originally conceived for addressing trans-radial amputations, because the extrinsic muscles
originally devoted to specific functions in the lost hand (e.g., the one responsible for flexing the fingers), and which are hosted inside the forearm, need to be targeted by the implant. This is because only by exploiting such native control sources we could realize an interface capable of restoring a natural control for the user. For this reason, after the search of general
guidelines in a simplified environment, the first part of the present dissertation is dedicated to the development and optimization of a multi-magnet localizer in a workspace resembling the human forearm, which is anatomically relevant for
the final application. By simulating different amputation levels, we determined the number of implantable magnets given the
stump length, as well as an optimal magnet placement inside the muscles that guarantees highly accurate localization outcomes. In addition, a sensor selection strategy aimed at reducing as much as possible the computational cost and power
consumption of the localization, while keeping the relevant information needed for localizing the magnets is presented.
On the other hand, the second part of this thesis tries to overcome the initial limit of application foreseen for the myokinetic approach, and to go beyond trans-radial amputations by addressing more proximal (above-the-elbow) ones, while
still preserving the expected naturalness of use of the proposed solution. This is made possible by the recent progress
in surgery, which has led to the so called targeted muscle reinnervation (TMR) technique: by redirecting the nerves that
served the lost limb to reinnervate surrogate muscles, such technique provides novel and physiologically appropriate control sources for artificial limbs. Thus, by merging the advances from surgery with those from technology, the viability and the
benefits of applying the myokinetic approach to proximal amputations in TMR patients were demonstrated. Specifically, experimental sessions carried out with two TMR patients proved the feasibility of applying the myokinetic interface, in a non
invasive way, to effectively decode different voluntary movements using a simple logistic regressor.
The outcomes of the present work not only are of interest for the development of a novel control interface for hand prostheses, but they also provide interesting insights for a wide range of bioengineering applications exploiting magnetic tracking.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|